Draft version May 21, 2018 A Preprint typeset using LTEX style emulateapj v. 05/12/14 PRELIMINARY EVIDENCE FOR A VIRIAL SHOCK AROUND THE COMA GALAXY CLUSTER Uri Keshet1, Doron Kushnir2, Abraham Loeb3, and Eli Waxman4 Draft version May 21, 2018 ABSTRACT Galaxy clusters, the largest gravitationally bound objects in the Universe, are thought to grow by accreting mass from their surroundings through large-scale virial shocks. Due to electron acceleration in such a shock, it should appear as a γ-ray, hard X-ray, and radio ring, elongated towards the large-scale filaments feeding the cluster, coincident with a cutoff in the thermal Sunyaev-Zel’dovich (SZ) signal. However, no such signature was found until now, and the very existence of cluster virial shocks has remained a theory. We find preliminary evidence for a large, ∼ 5 Mpc minor axis γ-ray ring around the Coma cluster, elongated towards the large scale filament connecting Coma and Abell 1367, detected at the nominal 2.7σ confidence level (5.1σ using control signal simulations). The γ-ray ring correlates both with a synchrotron signal and with the SZ cutoff, but not with Galactic tracers. The γ-ray and radio signatures agree with analytic and numerical predictions, if the shock deposits ∼ 1% of the thermal energy in relativistic electrons over a Hubble time, and ∼ 1% in magnetic fields. The implied inverse-Compton and synchrotron cumulative emission from similar shocks can significantly contribute to the diffuse extragalactic γ-ray and low frequency radio backgrounds. Our results, if confirmed, reveal the prolate structure of the hot gas in Coma, the feeding pattern of the cluster, and properties of the surrounding large scale voids and filaments. The anticipated detection of such shocks around other clusters would provide a powerful new cosmological probe. 1. INTRODUCTION However, no such virial shock signature has been de- In the hierarchical paradigm of large-scale structure tected until now, although a stacking analysis of EGRET (LSS) formation, galaxy clusters are the largest objects data around a sample of 447 rich clusters did sug- 14 gest a 3σ signal (Scharf & Mukherjee 2002). Upper ever to virialize. With masses in excess of 10 M⊙, they limit were imposed on the γ-ray emission from clus- are located at the nodes of the cosmic web, where they ters such as Coma (Sreekumar et al. 1996; Reimer et al. accrete matter from the surrounding voids and through 2003; Ackermann et al. 2010; Arlen et al. 2012, hence- large-scale filaments. With their vast size, galaxy clusters forth V12), including analyses of Coma based on Fermi resemble remote island universes, providing a powerful data (Ackermann et al. 2010, 2014), mostly focusing on cosmological probe and a unique astrophysical labora- the central parts of the cluster, well within the virial tory. radius; see discussion in §10.1. The very existence The gas accreted by a cluster is thought to abruptly of cluster-scale virial shocks has thus remained uncon- heat and slow down in a strong virial shock wave sur- firmed. rounding the cluster. Such collisionless shocks should, The Coma cluster (Abell 1656) is one of the richest by analogy with supernova remnant shocks, acceler- 15 ate charged particles to > TeV energies, where they nearby clusters. With mass M ∼ 10 M⊙, temperature Compton-scatter cosmic microwave-background (CMB) kBT ∼ 8 keV, and richness class 2, it lies only ∼ 100 Mpc away (Gavazzi et al. 2009), at a redshift z ≃ 0.023. The photons up to the γ-ray band (Loeb & Waxman 2000; ◦ Totani & Kitayama 2000; Keshet et al. 2003). Conse- cluster lies near the north Galactic pole (latitude ∼ 88 ), quently, one expects to find γ-ray rings around clus- in a sky patch remarkably devoid of Galactic foreground. These considerations, and indications for a high accre- arXiv:1210.1574v4 [astro-ph.CO] 22 Sep 2017 ters (Waxman & Loeb 2000), as indicated by cosmologi- cal simulations (Keshet et al. 2003; Miniati 2002), which tion rate as discussed below, render Coma exceptionally suggest an elliptic morphology elongated towards the suitable for the search for virial ring signatures. The virial radius of Coma, R ≃ R200 ≃ 2.3 Mpc, cor- large-scale filaments feeding the cluster (Keshet et al. ◦ 2003). Such rings are also expected in hard X- responds to an angular radius ψ ≃ ψ200 ≃ 1.3 . Here, rays (Kushnir & Waxman 2010), and should coincide subscripts 200 refer to an enclosed density 200 times with a synchrotron radio ring (Waxman & Loeb 2000; above the critical density of the Universe. The clus- Keshet et al. 2004a,b) and with a cutoff in the thermal ter is somewhat elongated in the east–west direction, in Sunyaev-Zel’dovich (SZ) signal (Kocsis et al. 2005). coincidence with the western LSS filament (West et al. 1995) that connects it with the cluster Abell 1367 (see [email protected] Figure 8). There is X-ray (Simionescu et al. 2013; 1 Physics Department, Ben-Gurion University of the Negev, Uchida et al. 2016), optical, weak lensing (Okabe et al. Be’er-Sheva 84105, Israel 2 2010, 2014), radio (Brown & Rudnick 2011), and SZ Institute for Advanced Study, Einstein Drive, Princeton, (Planck Collaboration et al. 2013) evidence that the New Jersey 08540, USA 3 Harvard-Smithsonian Center for Astrophysics, 60 Garden cluster is accreting clumpy matter and experiencing weak St., Cambridge, MA 02138, USA shocks towards the filament well within the virial radius, 4 Physics Faculty, Weizmann Institute of Science, POB 26, at ψ ∼ 0.5◦ radii. Rehovot, Israel 2 (a) VERITAS significance map (b) Simulated cluster in γ-rays Figure 1. Observed and simulated γ-ray maps of Coma (notice the different scales). Left: VERITAS & 220 GeV nominal significance map (V12) of Coma for θ = 0.2◦ integration (illustrated by the red central circle). Elliptical bins are shown (thin dashed cyan contours) for ∆b = 0.2◦, ζ ≡ a/b = 2 and φ = −5◦. The bins showing enhanced emission are highlighted (bounded by thick, long-dashed green curves). Right: Simulated map of a Coma-like cluster from a ΛCDM simulation (Keshet et al. 2003). The largest in the 200 Mpc simulation box, 15 ◦ this cluster has (Keshet et al. 2004a) mass M ≃ 10 M⊙ and temperature kB T ≃ 8 keV, like Coma. The 8.5 diameter image was convolved with a ∼ 0.23◦ beam, comparable to the VERITAS map, and rotated such that the large-scale filament extends to the west. −8 −2 −1 −1 Colorbar: log10(J/10 cm s ster ) brightness above 220 GeV for acceleration efficiency ξem˙ = 1%. The regions corresponding to the VERITAS mosaic (solid cyan contour) and to the VERITAS ring (elliptic dashed contours) are similarly highlighted. The VERITAS Cerenkovˇ array has produced a d ∼ 17%, and a hydrogen mass fraction χ =0.76. The plasma 4.8◦ diameter γ-ray mosaic (V12) of Coma, at energies is approximated as an ideal gas with an adiabatic index ǫ & 220 GeV. We argue that the significance map (Fig- Γ=5/3 and a mean particle massm ¯ = 0.59mp, where ure 1a) shows evidence for extended γ-ray emission away mp is the proton mass. from the center, that appears as a (∼ 0.6◦) thick elliptical ring with semi-minor axis b ≃ 1.3◦, elongated along the 2. VERITAS γ-RAY RING AROUND COMA east–west direction (best fit: φ ≃−5◦), with semi-major The VERITAS collaboration has presented (V12) a to semi-minor axes ratio ζ ≡ a/b & 3. The nominal sig- d ∼ 4.8◦ diameter γ-ray mosaic of Coma, at energies nificance of the signal is S =2.7σ, but accounting for the & 220 GeV. The significance map (Figure 1a) suggests background removal indicates a 5.1σ significance (for a some extended γ-ray emission away from the center. The blind search in b, ζ and φ). This preliminary signal agrees γ-ray structure appears as an elliptical ring, elongated with analytic and numerical predictions, and correlates along the east-west direction, or as two parallel east- with other tracers of the shock. west filaments lying symmetrically both north and south The paper is organized as follows. We analyze the of the cluster. The spectrum of the VERITAS feature is VERITAS mosaic in §2, present additional evidence for probably flat, as a p =2.4 photon spectral index was used extended γ-ray emission in §3, and show that other VER- to optimize the gamma-hadron separation cuts (V12). In ITAS fields show no such signal in §4. In §5 we present §8 we show that a flat spectrum with index p = 2 (equal the simulated VERITAS signature of a Coma-like virial energy per logarithmic energy interval, as expected in shock. The signal is shown to be unaffected by the Galac- the strong virial shock) is consistent with observations tic foreground in §6, but positively correlated with ex- at both lower and higher energies. pected radio signals in §7. We show in §8 that the γ-ray signal corresponds to an acceleration efficiency (over a 2.1. Foreground removal: ring model Hubble time) of ∼ 1%, and in §9 that the radio correla- Due to the strong foreground, the significance sj at a tions are consistent with ∼ 1% magnetization efficiency. given direction j in the VERITAS map is estimated by Finally, we summarize and discuss our results in §10. A comparing the numbers of events arriving from the tar- simple β-model for emission from the virial shock is given get vicinity (the so-called on region), Non, and away from in Appendix §A. it (the off region), Noff. These counts are then weighted We adopt a concordance flat ΛCDM model with Hub- by the detector acceptance, which varies as a function of −1 ble constant H = 70 km Mpc , a baryon fraction fb = the angular distance from the detector pointing.